This appendix contains an overview of the differences between Python 2 and Python 3, as far as they relate to the contents of this book and as far as I am aware of them. You can ignore this appendix if you are only going to use Python 3.
The division operator (/) in Python 2 works differently from the one in Python 3. In Python 3, it is automatically assumed that if you use division, you need floating-point numbers, so the division will assume that all numbers involved are floats, and will always result in a float. In Python 2, it is assumed that the division will be of the type that is “most detailed” for the numbers involved, i.e., if you divide two numbers and at least one is a float, it will be a floating-point division and the result is a float, but if you divide two integers, it will be an integer division and the result will be an integer (if the resulting number would mathematically have decimals, then the decimals are simply removed). The way Python 2 works is similar to what most programming languages do, but the way Python 3 works is more intuitive and leads to fewer errors.
In Python 3, print and exec are no longer reserved words (or
keywords), as they are now functions. However, True, False, and
None are now keywords, as is the word nonlocal.
A small difference between Python 2 and Python 3 is that when using the
type() function, where Python 3 displays the word class, Python 2
displays the word type. The reason is that in Python 3, all types are
actually implemented as classes.
The format() function was implemented in later versions of Python 2,
but did not exist in the earlier versions. Instead, it supported a style
of formatting, using “percentage-codes,” similar to what is used in
languages such as C++. This was directly implemented in the print()
function, and remained part of the print() function even after the
introduction of format() (for compatibility reasons). Consequently,
most Python 2 programs, even the ones written using the most recent
version of Python 2, use this older-style formatting. These older
approaches are no longer supported in Python 3. By the way, the
parentheses after the print() function are not necessary in Python 2,
but obligatory in Python 3.
In Python 3, the range() function is an iterator (see Chapter
24).
This entails that it needs very little memory space: it only retains the
last number generated, the step size, and the limit that it should
reach. In Python 2 range() is implemented differently: it produces all
the numbers of the range in memory at once. A statement like
range(1000000000) in Python 2 requires so much memory that it may very
well crash the program. In Python 3, such issues do not exist. In Python
2 it is therefore recommended not to use range() for more than 10,000
numbers or so, while in Python 3 no restrictions exist.
In Python 2, the string manipulation methods were part of the string
module, and were generally called by importing the module and then
writing string.<method>(). Such calls are no longer supported by
Python 3.
In Python 2, sorting a mixed list does not give a runtime error. The
sort() function also supports an argument cmp=<function>, that
allows you to specify a function that does the comparison between two
elements. This function is no longer supported in Python 3, but you can
use the key parameter for the same purpose. In the Python module
functools a function cmp_to_key() is found that turns the
old-style cmp specification into the new-style key specification.
In Python 2, the dictionary methods keys(), values(), and items()
return lists instead of iterators. Python 2 also supports a method
has_key() that you can use to check if a certain key is in a
dictionary, but this method has been removed from Python 3.
Python 2 does not support sets natively. You have to import the sets
module to use them. Moreover, to create a set you use the Set()
method, not the set() function. To create a set with elements in them,
in Python 2 the only way is to give the elements as a list argument to
the Set() method.
The structure of exceptions has changed a bit between Python 2 and
Python 3. In particular, the exception IOError has been made an alias
for the OSError exception, as it was quite hard to distinguish which
each of them tried to capture.
Python 2 was not yet supporting Unicode natively, while Python 3 does. Python 3 strings are Unicode strings. You will not notice the difference as long as strings are ASCII, but Python 2 strings might get into troubles when Unicode characters are inserted into the strings. Python 3 also supports Unicode characters in identifiers such as variable names, while Python 2 does not. It is not recommended that you do use non-ASCII characters in identifiers, though.
Python 2 does not have byte strings. There is no need for them in Python
2, as it does not support Unicode. In Python 2, if you want to write a
byte with ordinal value zero, you just write chr(0). The read() and
write() methods for binary files use regular strings in Python 2. This
is not allowed in Python 3.
In Python 2, the pickle module worked on text files. This is no longer
possible in Python 3, as Python 3 supports Unicode. Pickle files written
with Python 2 cannot be loaded in Python 3 and vice versa.
Python 3 is much more based on iterators and generators than Python 2, which has all kinds of advantages, mainly as far as speed and memory usage is concerned. Consequently, there are quite a few differences between Python 2 and 3 in this respect. I have not inventoried all of them, but here are a few that struck me:
Iterators in Python 2 have a next() method. They no longer have it in
Python 3, where it is called __next__().
In Python 2 zip() produces a list rather than an iterable.
The itertools module has some differences too. For instance, in Python
2 it has a function izip() that produces an iterable, but since in
Python 3 zip() does that already, the function izip() has been
removed from itertools.
Besides pickle and itertools, urllib has been changed considerably
between Python 2 and Python 3.
The statistics module does not exist for Python 2.